dx9 vs dx11 in farcry 3 huge performance difference, why is that?

[semperfidelis]

Hi,

I was just playing far cry 3 and i was testing some settings.

Weird thing is when playing on dx9 i got about 85% load on both gpu's. When playing on dx11 load is only about 35%. Why is this? I know dx11 is more efficient but seriously? 😱 And btw aa was turned off

 

[tadej petric]

New GPUs are better optimised for DX11 too.
Ooh and did you have AA disabled on both modes? AA is demanding as hell.

 

[Alex The PC Gamer]

This is a very simple question but there's no simple answer to this one. Strickly speaking in GPU language, both you guys are right. The same task/render in DX11 may be equal or better optimized generally speaking. That said, the libraries are different and DX11 has more options to render textures and geometry in general.

That's where the game engine comes into play and manages the main distribution of tasks (sends it to either GPU or CPU for calculations). Each engines operates differently so the Dunia Engine (an derivative engine from Crytek's CryEngine) will operate quite differently than the Unreal engine.

Furthermore, the graphics drivers will maximize GPU performance (based on the application).

So your results could either be 2 things...

1 - The way the Engine/drivers handle DX9 and DX11 calculations are very different...perhaps more CPU intensive when it comes to DX11, which i doubt is the case.

2 - Or your results aren't comparable...benchmarks need to be replicable (or very close to identical), otherwise you'll end up with biased data.

It'd be interesting to see how the CPU load behaves between the same benchmark with that engine (and comparing DX9 to DX11).

That's where graphics drivers come in and help maximize performance

 

[semperfidelis]

New GPUs are better optimised for DX11 too.
Ooh and did you have AA disabled on both modes? AA is demanding as hell.

Settings were identical. The ingame AA causes severe stuttering for me.

 

[semperfidelis]

This is a very simple question but there's no simple answer to this one. Strickly speaking in GPU language, both you guys are right. The same task/render in DX11 may be equal or better optimized generally speaking. That said, the libraries are different and DX11 has more options to render textures and geometry in general.

That's where the game engine comes into play and manages the main distribution of tasks (sends it to either GPU or CPU for calculations). Each engines operates differently so the Dunia Engine (an derivative engine from Crytek's CryEngine) will operate quite differently than the Unreal engine.

Furthermore, the graphics drivers will maximize GPU performance (based on the application).

So your results could either be 2 things...

1 - The way the Engine/drivers handle DX9 and DX11 calculations are very different...perhaps more CPU intensive when it comes to DX11, which i doubt is the case.

2 - Or your results aren't comparable...benchmarks need to be replicable (or very close to identical), otherwise you'll end up with biased data.

It'd be interesting to see how the CPU load behaves between the same benchmark with that engine (and comparing DX9 to DX11).

That's where graphics drivers come in and help maximize performance

I didnt really monitor my CPU since this game seems to run very smooth on ultra and supersampling AA turned on through CCC.
Only the Postfx I have to turn of cause it causes a massive fps-drop (40+) when looking over the ocean/sky. Which is kind of weird because there is realy not much to see there...

 

[ajay272]

Their is huge difference in dx9 vs dx11 ... In dx11 mode game have... Soft shadow and higher quality shaders... That's why DX 11 have lil more demanding then dx9... And u saying in dx11 your GPU using lesser power then DX 9 .. But on 2560X1440p it using maximum power of my gtx 970.. And giving around 55-65 fps In dx11...